The dementia of Alzheimer's disease (AD) follows years of pre-symptomatic neurodegeneration. An ideal AD intervention would therefore reduce the occurrence of dementia by slowing or stopping the disease process in its pre-symptomatic stages. Laboratory results suggest that prostaglandin-dependent inflammatory or signaling mechanisms may be important in AD pathogenesis, and epidemiologic evidence suggests that conventional NSAIDs that inhibit COX-1 or COX-2 activation of the prostaglandin pathway can reduce AD dementia incidence. The masked, placebo-controlled Alzheimer's Disease Anti-inflammatory Prevention Trial (ADAPT) was therefore designed to test whether the dual-inhibitor NSAID naproxen or the COX-2 selective agent celecoxib can reduce the incidence of AD dementia and mitigate age-related cognitive decline. Expectations for informative results from ADAPT were diminished when its treatments were stopped after an average 21 months of treatment assignment. Indeed, recently published results from ADAPT showed no benefit from NSAID exposure, but instead suggested an early increase in the incidence of dementia with treatment. Over the past 22 months, however, the naproxen-assigned group has shown reduced AD dementia incidence, and the contrast between the earlier and later direction of effects with naproxen is striking. Unfortunately, interruption of funding has necessitated an end of data gathering and the unmasking of ADAPT just when its predicted treatment effects seem to be emerging. This application therefore proposes to follow up the ADAPT cohort over an interval of 30 months, using the telephone to screen for cognitive difficulty and then evaluating screen-positive individuals for new dementia outcomes. We will then repeat this process a year later and combine the new results with those of ADAPT to accrue some 180 total AD dementia "events" as well as an extensive array of longitudinal cognitive test scores. Our Specific Aim 1 is thus to extend observations on the ADAPT outcomes of incident AD dementia and age-related cognitive decline in relation to treatment assignment. We seek in particular to discover whether naproxen can cause a long-term reduction in the incidence of AD dementia. Also (as permitted by the data) we seek to characterize a "bi-directional" or "bi-phasic" response of elderly people to naproxen (or, if any, to celecoxib) with a brief interval of increased incidence followed by a more substantial epoch of diminished risk. Specific Aim 2 is to determine whether the limited exposures in ADAPT to either of the two NSAID treatments produced long-term change in age-related cognitive decline as a possible early manifestation of AD pathology. Although safety concerns will probably negate the utility of current generation NSAIDs for prevention of AD dementia, the elucidation of their effects on the disease in humans is of great importance as an indicator of AD pathogenetic mechanisms and future prevention strategies.